Adding synthetic MM to a FSL basis set

MRS Software FSL-MRS
1

I am trying to use FSL to do a comparison between FSL and Osprey MRS and fMRS processing. We did not collect experimental macromolecule data for our basis set, so I would like to add it into my basis set using the FSL command: “basis_tools add_set --add_MM”. I was successfully able to convert the LCModel .basis to an FSL basis set of .json files.

When I run the command basis_tools add_set --add_MM, the basis set does not visually change, but there are now .json files for each MM in the basis set.

However, when I run the fit with the basis set with added MM, the fit doesn’t work. I assume the fit fails because I am asking it to fit MM that didn’t get correctly added to the basis set.

Regardless, does anyone have suggestions for how to properly add MMs, or why the fit is failing otherwise?
Thank you for any input you can provide!

The following is the fitting call that I have been using:
‘fsl_mrs’,
‘–data’, data_path,
‘–basis’, ‘basis_MM’,
‘–output’, output_path,
‘–h2o’, wref_path,
‘–baseline’, ‘spline, moderate’,
‘–metab_groups’, ‘MM09’, ‘MM12’, ‘MM14’, ‘MM17’, ‘MM21’,
‘–TE’, ‘5’,
‘–TR’, ‘10’,
‘–report’,
‘–overwrite’)

I added the “metab_groups” line only when using the basis with added MM.

I can only add one embed. The fit without MM is a typical fit. The fit WITH MM is nonsense:

fit_summary
fit_summary900×1000 62.8 KB

2

Hi @AlexB,

I think this is a frequency convention thing. If you run

basis_tools vis --ppmlim -20 20 path/to/basis_with_mm_added

Do you see the peaks on the wrong side of water?

If so first run

basis_tools conj path/to/basis path/to/basis_conj

then

basis_tools add_set --add_MM path/to/basis_conj

Hope that fixes the issue.

Will

3

Hi Will, thanks for the response.

That was exactly it. This is my new basis after doing those steps:

conj_before_add
conj_before_add640×480 24.3 KB

The fit works correctly now, but this doesn’t look quite right to me. Why is the basis set shifted around 6 ppm in this image? Shouldn’t it be in the 0.8-4.2 ppm range? Or am I misunderstanding what this visualization is?

Thanks again for the help,

Alex

4

Nice.

The frequency/phase convention for basis sets between toolboxes (and, as you have found, to a within FSL) is inconsistent. This is sorted automatically in the fsl_mrs call to fit, but you could also sort it manually by running

basis_tools conj path/to/basis_conj path/to/basis_final